fabric-dataflowlisted
Install: claude install-skill wardawgmalvicious/agent-config
# Fabric Dataflow Gen2 (the `Dataflow` item)
Dataflow Gen2 is the Power Query item in Fabric Data Factory: a low-code
transformation surface that ingests, reshapes and loads data without
code. This skill covers the item as **Git and REST see it** — the files
on disk, the API that writes them, and the limits that decide whether a
design works.
**The product name and the item type differ, and both are load-bearing.**
Everything a user types says *Dataflow Gen2*; the `.platform` file, the
`ItemType` enum and the REST namespace all say **`Dataflow`**. The Git
folder is `<display name>.Dataflow`.
**"Gen2" no longer implies a choice.** Since April 2026 the option to
create a Dataflow Gen2 *without* CI/CD and Git support is gone, and every
new item is created with it. Items created before that keep working, and
the docs still split behaviour by which kind you have — so when a page
says "applies to Dataflow Gen2 with CI/CD support", that is now the
default case and the non-CI/CD case is legacy.
## 1. What is on disk
| File | Required | Contents |
| --- | --- | --- |
| `mashup.pq` | yes | The Power Query document — every query, as M |
| `queryMetadata.json` | yes | Query options: load flags, groups, connections, compute settings |
| `<name>.mdf` | no | A Mapping Data Flow transform; there can be several |
| `.platform` | — | Item metadata — `displayName`, `logicalId` |
Serialization mechanics — EOF newlines, CRLF/LF, `.gitattributes` — are
in the `fabric-git-serialization` rule,